search

YongyuG / rnnoise_16k

implementation of rnnoise_16k
BSD 3-Clause "New" or "Revised" License
123 stars 40 forks source link

can i use 128 as framesize when using model , if not, how can I do it ? #18

Open zhuhao528 opened 1 year ago

zhuhao528 commented 1 year ago

我可以在使用模型的时候使用帧长度为128吗，我看现有的模型是使用的帧长度是160。如果在使用模型的时候使用128帧长不合适，那么需要改网络训练的帧长，重新训练吗？

YongyuG commented 1 year ago

you can use any framesize you want, set it in denoise.c macro definition

发件人: zhuhao528 @.> 发送时间: 2023年8月30日 9:43 收件人: YongyuG/rnnoise_16k @.> 抄送: Subscribed @.***> 主题: [YongyuG/rnnoise_16k] can i use 128 as framesize when using model , if not, how can I do it ? (Issue #18)

― Reply to this email directly, view it on GitHubhttps://github.com/YongyuG/rnnoise_16k/issues/18, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AH7IVVVDC7AQVE6WALJNJ5LXX34MNANCNFSM6AAAAAA4EE6CS4. You are receiving this because you are subscribed to this thread.Message ID: @.***>

zhuhao528 commented 1 year ago

ok thanks，I read a blog that helped me，this is the address https://codeantenna.com/a/pEQArF4zmR

zhuhao528 commented 1 year ago

when modifing the frame_size to 128, should i modify eband5ms array as same time? for example: /for 16K speech frame_size 160 files/ // hz[i]/4(sample_rate/win_len) = hz[i]/200 // static const opus_int16 eband5ms[] = { ///0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 20k*/ // 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 34, 40, 48, 60, 78, 100 //};

/for 16K speech frame_size 128 files/ // hz[i]/4(sample_rate/win_len) = hz[i]/250 向下取整 static const opus_int16 eband5ms[] = { /0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 20k*/ 0, 0, 1, 2, 3, 4, 4, 5, 6, 8, 9, 11, 12, 16, 19, 22, 27, 32, 38, 48, 62, 80 };

I round up when I calculate，is this right？Frequency range for boys is 100-200hz，girl is 200-350Hz，will this have a detrimental effect？

YongyuG commented 1 year ago

when modifing the frame_size to 128, should i modify eband5ms array as same time? for example: /_for 16K speech framesize 160 files/ // hz[i]/4*(sample_rate/win_len) = hz[i]/200 // static const opus_int16 eband5ms[] = { ///0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 20k/ // 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 34, 40, 48, 60, 78, 100 //};

/_for 16K speech framesize 128 files/ // hz[i]/4*(sample_rate/win_len) = hz[i]/250 向下取整 static const opus_int16 eband5ms[] = { /0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 20k/ 0, 0, 1, 2, 3, 4, 4, 5, 6, 8, 9, 11, 12, 16, 19, 22, 27, 32, 38, 48, 62, 80 };

I round up when I calculate，is this right？Frequency range for boys is 100-200hz，girl is 200-350Hz，will this have a detrimental effect？

when modifing the frame_size to 128, should i modify eband5ms array as same time? for example: /_for 16K speech framesize 160 files/ // hz[i]/4*(sample_rate/win_len) = hz[i]/200 // static const opus_int16 eband5ms[] = { ///0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 20k/ // 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 34, 40, 48, 60, 78, 100 //};

/_for 16K speech framesize 128 files/ // hz[i]/4*(sample_rate/win_len) = hz[i]/250 向下取整 static const opus_int16 eband5ms[] = { /0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 20k/ 0, 0, 1, 2, 3, 4, 4, 5, 6, 8, 9, 11, 12, 16, 19, 22, 27, 32, 38, 48, 62, 80 };

I round up when I calculate，is this right？Frequency range for boys is 100-200hz，girl is 200-350Hz，will this have a detrimental effect？

no need actually, but you can do it as a experimentt.